Toward Understanding the Behavior of BGP During Large-Scale Power Outages
|
|
- Kristian Cummings
- 5 years ago
- Views:
Transcription
1 Toward Understanding the Behavior of BGP During Large-Scale Power Outages Jun Li, Zhen Wu, and Eric Purpus {lijun, zwu, Computer and Information Science Department University of Oregon Abstract While the Internet continues to thrive, the resiliency of its fundamental routing infrastructure is not fully understood. In this paper, we analyze the behavior of the de facto inter-domain routing protocol, BGP, during a large-scale power outage that affected the connectivity of 3, 175 networks in dozens of cities in the eastern USA and Canada. By observing proper metrics of BGP, we study BGP behavior from both the global level and the prefix level. At the global level, our results show that many global BGP metrics remained stable during the blackout event; importantly, we do not find an increase in the number of BGP announcements, a metric that has been primarily used to indicate significant changes in routing. However, we observe an apparent increase in the number of withdrawals. At the prefix level, we introduce per-prefix AS-path graphs and study their evolution for affected prefixes during the blackout; we have found that in such graphs there is a sharp decrease in the number of edges and nodes as well as changes in node degrees. I. INTRODUCTION As the Border Gateway Protocol (BGP) [1] provides the inter-domain routing throughout the Internet a fundamental functionality for delivering Internet packets across different domains, failure or degradation of BGP can lead to various severe problems, ranging from packet loss or delay to the collapse of the whole Internet. It is therefore critical to understand how BGP reacts to adversity and how resilient BGP is. As there are many different kinds of adversity, such as largescale power outages, misconfiguration, and worm attacks, in this paper we study BGP under specific events; in particular, we choose the East Coast blackout of August 23. Our objective is to characterize and analyze BGP behavior surrounding this event, and try to deduce the resiliency and responsiveness of BGP during such a large-scale adversity. We are particularly interested in discovering any primary distinction of BGP behavior during this blackout. The East Coast blackout is reportedly the largest of its scale and affected the connectivity of 3,175 networks [2]. Shortly after 4 p.m. EDT on August 14, 23, this power outage hit a large region in the eastern USA and Canada within two minutes. It was not until 9 p.m. EDT on August 15 that power was completely restored. More details of the outage, such as how networks of various sizes were affected, can be found in [2]. In this paper, we analyze BGP behavior mainly from two complimentary angles: the global-level BGP behavior analysis and the prefix-level analysis. We believe combining the results at these two levels will strengthen our analysis. Global level analysis may reveal the existence of anomalies or differences in BGP behavior. Prefix level analysis may reveal details about the nature of an anomaly or difference. Furthermore, unusual behavior at the prefix level may be hidden at the global level by the noise of the global statistics. Our paper makes the following contributions. First, we propose several global metrics to characterize macroscopic BGP behavior. We discover that although no significant difference is observed in the number of total BGP updates or other metrics, the number of explicit withdrawal messages makes the blackout stand out. Second, we introduce per-prefix AS-path graphs to study BGP behavior from a microscopic point of view, and our results demonstrate that this is a promising approach to observe prefix-level routing changes or diagnose routing problems. II. METHODOLOGY A. Overview Using BGP update data collected before, during, and after the blackout event, we gather BGP statistics at both the global and prefix level. At the global level, we gather statistics in terms of several representative metrics on BGP updates: the number of BGP updates, the number of explicit BGP withdrawals, interval of BGP updates, and top active prefixes. At the prefix level, we create graphs describing the routing paths toward a destination prefix AS from other AS sources in the Internet, i.e. per-prefix AS-path graphs, and examine the changes in these paths throughout the blackout event. For both global-level and prefix-level analysis, we use the data before and after the blackout event to serve as a baseline of assumed normal behavior. Our graphs will have three vertical time-marking lines: the first line marking the beginning of the power outage on August 14, 2:1 UTC, the second the beginning of the incremental power recovery process on August 15, 4: UTC, and the third the end of the process on August 16, 2:3 UTC. B. BGP Updates BGP routers exchange updates between each other to update the path information for particular prefixes. Updates come in the form of either announcements or withdrawals, where paths are added or removed respectively. Each update is relevant to a list of prefixes that are included in the update. In announcements,
2 each prefix updates include an AS-path to describe the path taken to reach that prefix. Our study uses the BGP updates from the University of Oregon RouteViews archive [3]. By peering with ASes from around the world, RouteViews monitors continuously collect BGP updates from their BGP peers (those collected during a table dumping process are filtered in our study), resulting in a huge archive of BGP updates regarding reaching prefixes throughout the whole Internet. C. Per-Prefix AS-path Graphs We construct a per-prefix AS-path graph as follows. Starting with an empty graph, we filter all the updates for a particular prefix during a given window of time. For every announcement, we add the announced path to the graph if it does not already exist, based on the AS-path field for this prefix. The routerid of the announcement is also associated with that path to differentiate between multiple routers from the same AS. For every withdrawal, we match the router-id of the withdrawal with the path in the graph associated with the same router-id, and remove this path from the graph. We use the router-id to do this because the AS-path being withdrawn is not explicitly given, but instead is implied in the withdrawal. At this point, any unconnected AS nodes are removed from the graph. III. GLOBAL-LEVEL STATISTICS Since there is already a huge amount of BGP updates propagating throughout the Internet, and only a small amount of BGP updates would be produced for those networks affected by the blackout event, one would expect that hardly any statistical difference of BGP can be observed. This is true as we analyze the results on the number of BGP updates, the interval of BGP updates, and top active prefixes. However, we do observe an apparent increase in the number of explicit BGP withdrawals. Note that because the amount of explicit withdrawals is only about 5% of all BGP updates, changes regarding explicit withdrawals are easier to notice. A. Number of BGP Updates The number of BGP updates has been a primary metric in observing the dynamics of BGP. Changes in the number of updates over a certain period directly indicates whether BGP peers are relatively chatty or quiet, or even perhaps experiencing some difficulties in exchanging routing information. Previous major events, such as Internet worms, often caused obvious surges in the number of BGP updates ([4], [5]). Naturally, we want to answer the same question: Did the blackout cause a similar surge in the number of BGP updates? We observe no significant changes in the number of BGP updates during or after the blackout. Figure 1 shows the number of updates received by our RouteViews monitor every hour. Although there is an increase immediately after the power outage, it does not stand out compared to the number of updates when the blackout event is not happening. This is not surprising considering that a power outage is an ON/OFF event. B. Interval of BGP updates The interval between BGP updates (also called inter-arrival time) can also indicate the level of BGP activity and the general stability of routing paths. When no changes occur to the routing paths, no updates will be exchanged. Conversely, frequent path changes can lead to frequent routing information exchange through BGP updates, thus leading to smaller update intervals. In our measurement, we define the update interval as the time between two updates for the same prefix from the same RouteViews peer. We analyze the interval distribution during the blackout period and compare it to two randomly chosen periods of the same length. As shown in Figure 2, we found no significant difference. There are three spikes for all three periods, around interval values of 3, 6, and 9 seconds, and most intervals (about 7-8%) are less than 1 seconds. Both the blackout period distribution and the reference distributions have these spikes, and do not differ significantly from one another. We believe the spikes are the effect of a configuration parameter the Minimal Route Advertisement Interval (MRAI) timer that was introduced to control BGP traffic overhead on routers by defining the minimal amount of time between advertisements from a single router for a particular prefix. C. Top Active Prefixes Due to the ON/OFF nature of the power outage, we suspect that during the blackout period those affected prefixes were not significantly more active than those unaffected prefixes. To evaluate our hypothesis, we study the hourly rate of BGP updates for the top 1, active prefixes. As shown in Figure 3, we do not find significant variations during the blackout period. Although higher values exist immediately after the power outage, this less than significant variation can hardly be regarded as correlated with the blackout event. Additionally, our analysis has shown that the top 1, prefixes, which constitute less than 1% of the total number of unique prefixes in a routing table, typically contribute almost 4% of the total number of prefix occurrences every hour. And this distribution pattern did not change during the blackout. Finally, our analysis on the top 5 and 1 active prefixes also show that no significant differences can be observed during the blackout. Update Number Fig. 1. Date of August 23 Number of BGP updates every hour
3 Occurrence number Occurrence Number Blackout (Aug :1: - Aug :3:) Baseline (Aug :1: - Aug :3:) Baseline (Aug :1: - Aug :3:) Inter-arrival time (in seconds) Fig. 2. Update interval distribution Top 1 prefixes 1 Date of August 23 Fig. 3. Top active 1 prefixes D. Number of Explicit Withdrawals We now study the metric that indeed experienced significant variation during the blackout the number of explicit withdrawals. Withdrawals happen when a given path for reaching a particular prefix is no longer preferred or available. While a BGP peer can announce a different path to reach a prefix, also called an implicit withdrawal, the BGP peer can simply inform others through an explicit withdrawal that the path toward a given prefix is no longer valid. Although power backup systems are common, the power outage caused some networks to become unreachable due to the sustained nature of the outage, extending beyond backup capabilities. BGP sessions that involve networks affected by the blackout can thus be broken. When BGP routers that peer with BGP routers from affected networks send out explicit withdrawals to notify other BGP routers, it will result in many Withdrawal Number Date of August 23 Fig. 4. Number of BGP withdrawals every hour withdrawal updates. The number of networks affected during the blackout directly affects the number of withdrawal updates. In the case of the East Coast blackout, as shown in Figure 4, a spike of almost 2% in the number of withdrawals occurred immediately after the power outage. IV. PREFIX-LEVEL STATISTICS Global-level BGP analysis can show certain changes during a large-scale failure such as the East coast blackout event, but such an analysis is aggregated over all networks. In order to pinpoint problematic prefixes or perform any prefix-level operations, it is important to zoom in and watch BGP behavior at the individual prefix level. We construct prefix graphs every five minutes before, during, and after the blackout, and record their characteristics at these points. In the following, we first present an example per-prefix AS-path graph, then describe what metrics and which prefixes to use for analysis, and finally present our results and analysis at the prefix level. A. An Example Per-Prefix AS-Path Graph Figure 5 shows an example of a per-prefix AS-path graph. The shaded nodes in the graph are ASes which contain a router that our RouteViews monitor directly peers with. The unshaded nodes represent ASes which we learn about from AS paths, but do not peer with. This graph has a number of interesting characteristics: AS 1221 is the origin AS, and contains routing loops to itself. These loops are formed when the AS path contains the AS number multiple times in a row. This is commonly known as prepending the origin AS to the AS path, and is often used as a method of traffic engineering for creating longer AS paths. AS 313 announces two different paths one through AS 2914, and one through AS Since each BGP router will only announce one path toward a prefix, we can conclude that there are actually two different routers from AS 313 announcing two different paths. We can identify this behavior by tracking the out-degree in the per-prefix graphs. In this case, the node for AS 313 has an out-degree of two. AS 1221 has four peers carrying its traffic toward this particular prefix, and two of those four peers have a larger number of peers routing traffic through them. Here we can characterize this kind of connectivity by in-degree. AS 71 has an in-degree of five, while AS 1529 has an in-degree of seven. B. What Metrics, and Which Prefixes? To analyze prefix-level BGP behavior during the blackout event, we examine the changes of per-prefix graphs throughout the blackout event. There are two issues to consider: what metrics to use to discover graph changes, and which particular prefixes to investigate. Per-prefix graphs present us with a multitude of powerful methods for analyzing graph evolution. We start with the following graph characteristics: number of nodes, number of edges, node in-degree, and node out-degree. The number of
4 Fig. 5. An example per-prefix AS-path graph nodes and edges can reflect the scale of the graph. Node degrees can indicate the connectivity of the graph. Variations in node degrees thus will reflect changes regarding how nodes are connected at different times. Note that here in and out are relative to the direction of the graph oriented towards the origin AS for a given prefix. In a graph where nodes have high in-degree, many ASes choose paths through a common AS. In a graph where nodes have out-degrees greater than one, multiple routers for a single AS choose different paths to reach a given prefix (particularly common for large ASes that span large regions). Furthermore, we observe that compared to the total number of ASes in a graph, the number of ASes that advertise different routes from more than one router is small. This makes the out-degree of a majority of the nodes exactly one, so in the following we actually do not inspect node out-degree. Meanwhile, with the out-degrees mostly one, the shape of the graph is very tree-like and the number of edges and nodes in the graph are directly related to one another (if it is exactly a tree, the number of nodes is equal to the number of edges plus one). Adding or removing an edge from the graph often requires adding or removing a node as well. Therefore, the number of nodes and edges over time will be almost identical in shape, and we leave out the number of edges in the following. We analyzed 63 prefixes with our per-prefix graph method. 17 of these prefixes were chosen randomly from ASes which are affected by the blackout according to the Renesys report [2]. The other 46 came from distinct ASes chosen at random. In the following, we use a prefix from an AS affected by the blackout as a representative to report our results and analysis. C. Number of Nodes Figure 6 shows the change in the number of nodes at five minute intervals. In particular, we see a drop in the number of nodes between the initial outage and the start of the power restoration. The reason of the drop is as follows. When a BGP router notices that a path to a particular prefix is no longer available, such as due to the unreachability of its next-hop BGP router, it will send out a withdrawal update to its BGP peers to which it has advertised the path. These peers will then further send out withdrawals of paths to this prefix. This phenomenon happens during the blackout when BGP speakers of affected networks become unavailable, sometimes after their power backup runs out. In managing the prefix graph for a Number of Nodes Fig : 2: 21: 22: 23: : 1: 2: Time (August 14-15) Number of nodes during the time window particular prefix, every withdrawal of a path toward this prefix will cause a corresponding path to be removed from the graph. After receiving enough withdrawals, there may be no paths to the prefix in question which include a given node. That node will then also be removed from the graph. A decrease in the number of nodes thus indicates that a number of withdrawals for this prefix have occurred. We can also see the recovery shortly after the drop. When a path towards a prefix becomes available again, the path is reannounced and propagates in a fashion similar to withdrawals. In managing the prefix graph, when announcements arrive for paths that do not already exist in the graph, the paths are then added to the graph. Figure 6 shows that the paths are quickly re-announced and the number of nodes remains fairly steady thereafter. D. Node In-Degree We look at the distribution of node in-degrees during the blackout period, including the average and maximum of their values. For a given graph of a particular prefix, the average is useful for observing overall graph connectivity, while the maximum shows the in-degree of the most popular AS, i.e., the AS with the greatest number of upstream ASes for reaching the prefix. Figure 7 shows the variation of the maximum in-degree during the blackout: a sudden increase followed by a sudden decrease and then a return to the original value. These two sudden changes reflect the changes in connectivity caused by the blackout. Suddenly, 15 ASes all switch to paths that have a common AS as the next hop, but then as the total number of possible paths drops, the maximum in-degree also drops.
5 Max In Degree Fig : 2: 21: 22: 23: : 1: 2: Time (August 14-15) Maximum in-degree during the East Coast blackout The first spike in in-degree indicates that many BGP routers had that particular common AS as a secondary route. When the preferred routes became unavailable to these routers, many of them switched to the same route through this common AS. Figure 7 also has a recovery phase. When the blackout ends and the graph recovers to its original level of scale and connectivity shape, the maximum in-degree also returns to its original value. V. RELATED WORKS The literature contains a number of studies on generic behavior and dynamics of BGP (e.g. [6], [7], [8]). Regarding the impact of the East Coast blackout on the Internet, Cowie et al. [2] found that 3175 networks were affected by the blackout, with over 2 networks suffering severe outage for more than 4 hours and over 14 networks for longer than 12 hours. BGP during a blackout shares commonalities with other studies on BGP under stress. These studies, including [4], [9], [1], [5], have focused on analyzing the global behavior of BGP, and the stresses are mainly from Internet worms. Similar to our study using per-prefix AS-path graphs, some works have also tried to study BGP from a microscopic point of view. An earlier work [11] by Govindan et al. studied the routing stability in the context of prefix reachability: prefix availability and prefix steadiness. Researchers also studied the reliability of Internet paths toward a sample set of prefixes [12]. Also similar to our approach to visualizing the detailed routing toward a prefix, another independent work by [13] also proposed to construct AS-level graph representations for reaching a given prefix, which is mainly a visualization tool. Finally, anomalies in reaching particular prefixes have also been studied [14]. VI. CONCLUSIONS While the Internet continues to thrive, BGP as the de facto inter-domain routing protocol over the Internet must be studied, especially when it is facing serious failures or attacks. In this paper, we analyzed BGP s behavior during a large-scale power outage that caused 3, 175 networks to lose their connectivity. We identified a number of metrics for analyzing BGP behavior, including both global level and prefix level, and applied these metrics to our analysis. Global metrics studied include number of updates, inter-arrival time, top active prefixes over a time window of two weeks surrounding the blackout period, and explicit withdrawals. We also introduced per-prefix AS-path graphs as another analysis method, and studied changes in graph characteristics such as the number of nodes and node indegrees. Our results show that during the blackout event, there was an apparent increase in the number of explicit withdrawals, and at the prefix level, the per-prefix AS-path graphs of some prefixes also show a sharp decrease in the number of edges and nodes, as well as changes in node in-degrees. Nonetheless, our results also show that BGP can recover from the blackout event in a timely manner. Overall, the negative impact of the power outage was limited only to the affected areas, and BGP performed well during this large-scale power outage. We also hope that our BGP behavior analysis techniques in this blackout event can be generalized and applied towards analyzing BGP s resiliency in other security, misconfiguration, or outage events. REFERENCES [1] Y. Rekhter, T. Li, and S. Hares. A Border Gateway Protocol 4 (BGP-4). RFC 4271 (Draft Standard), January 26. [2] J. Cowie, A. Ogielski, B. Premore, E. Smith, and T. Underwood. Impact of the 23 blackouts on Internet communications, November 23. [3] University of Oregon Route Views Project. RouteViews. antc.uoregon.edu/route-views/. [4] J. Cowie, A. Ogielski, B. Premore, and Y. Yuan. Global routing instabilities during Code Red II and Nimda worm propagation. Technical report, Renesys, 21. [5] M. Lad, X. Zhao, B. Zhang, D. Massey, and L. Zhang. An analysis of BGP update surge during Slammer attack. In Proceedings of the International Workshop on Distributed Computing (IWDC), 23. [6] C. Labovitz, G. Malan, and F. Jahanian. Internet routing instability. IEEE/ACM Transactions on Networking, 6(5): , [7] C. Labovitz, A. Ahuja, A. Bose, and F. Jahanian. Delayed Internet routing convergence. In Proceedings of ACM SIGCOMM, pages , 2. [8] T. Griffin and B. Premore. An experimental analysis of BGP convergence time. In Proceedings of the Ninth International Conference on Network Protocols (ICNP 1), pages 53 61, November 21. [9] J. Cowie, A. Ogielski, B. Premore, and Y. Yuan. Internet worms and global routing instabilities. In Proceedings of SPIE International symposium on Convergence of IT and Communication, 22. [1] L. Wang, X. Zhao, D. Pei, R. Bush, D. Massey, A. Mankin, S. Wu, and L. Zhang. Observation and analysis of BGP behavior under stress. In Proceedings of Internet Measurement Workshop, November 22. [11] R. Govindan and A. Reddy. An analysis of Internet inter-domain topology and route stability. In Proceedings of IEEE INFOCOMM, April [12] X. Zhao, M. Lad, D. Pei, L. Wang, D. Massey, S. Wu, and L. Zhang. Understanding BGP behavior through a study of DoD prefixes. In Proceedings of DISCEX III, April 23. [13] L. Colitti, G. Battista, I. Marinis, F. Mariani, M. Pizzonia, and M. Patrignani. BGPlay. [14] Ke Zhang, Amy Yen, X. Zhao, D. Massey, Felix S. Wu, and L. Zhang. On detection of anomalous routing dynamics in BGP. In Proceedings of the International IFIP-TC6 Networking Conference, pages , 24.
On the Impact of Route Processing and MRAI Timers on BGP Convergence Times
On the Impact of Route Processing and MRAI Timers on BGP Convergence Times Shivani Deshpande and Biplab Sikdar Department of ECSE, Rensselaer Polytechnic Institute, Troy, NY 12180 Abstract Fast convergence
More informationMeasurement of Highly Active Prefixes in BGP
1 Measurement of Highly Active Prefixes in BGP Ricardo V. Oliveira, Rafit Izhak-Ratzin, Beichuan Zhang, Lixia Zhang Abstract We conduct a systematic study on the pervasiveness and persistency of one specific
More informationUnderstanding BGP Behavior through a Study of DoD Prefixes Λ
Understanding BGP Behavior through a Study of DoD Prefixes Λ Xiaoliang Zhao, Dan Massey y USC/ISI S. Felix Wu z UC Davis Mohit Lad, Dan Pei, Lan Wang, Lixia Zhang x UCLA Abstract BGP is the de-facto inter-domain
More informationHappy Packets: Some Initial Results
Happy Packets: Some Initial Results RIPE / Manchester, UK 2004.09.22 Randy Bush Timothy G. Griffin Z. Morley Mao Eric Purpus
More informationInvestigating occurrence of duplicate updates in BGP announcements
Investigating occurrence of duplicate updates in BGP announcements Jong Han Park 1, Dan Jen 1, Mohit Lad 2, Shane Amante 3, Danny McPherson 4, and Lixia Zhang 1 1 University of California, Los Angeles
More informations i g c o m m Public Review for BGP Routing Dynamics Revisited Jun Li, Michael Guidero, Zhen Wu, Eric Purpus, and Toby Ehrenkranz
a c m Public Review for BGP Routing Dynamics Revisited Jun Li, Michael Guidero, Zhen Wu, Eric Purpus, and Toby Ehrenkranz BGP is indeed a critical protocol of the Internet. Understanding its behaviour,
More informationOn the Impact of Filters on Analyzing Prefix Reachability in the Internet. Ravish Khosla, Sonia Fahmy, Y. Charlie Hu Purdue University ICCCN 2009
On the Impact of Filters on Analyzing Prefix Reachability in the Internet Ravish Khosla, Sonia Fahmy, Y. Charlie Hu Purdue University ICCCN 2009 Background Border Gateway Protocol (BGP) Inter-domain policy
More informationWatching Data Streams Toward a Multi-Homed Sink Under Routing Changes Introduced by a BGP Beacon
Watching Data Streams Toward a Multi-Homed Sink Under Routing Changes Introduced by a BGP Beacon Jun Li, Randy Bush, Zhuoqing Mao, Timothy Griffin, Matthew Roughan, Daniel Stutzbach, Eric Purpus Abstract.
More informationDetection of Interdomain Routing Anomalies Based on Higher-Order Path Analysis
Detection of Interdomain Routing Anomalies Based on Higher-Order Path Analysis Murat Can Ganiz, Sudhan Kanitkar and Mooi Choo Chuah Lehigh University CSE Department Bethlehem, PA 18015, USA {mug3, sgk205,
More informationBGP Routing: A study at Large Time Scale
BGP Routing: A study at Large Time Scale Georgos Siganos U.C. Riverside Dept. of Comp. Science siganos@cs.ucr.edu Michalis Faloutsos U.C. Riverside Dept. of Comp. Science michalis@cs.ucr.edu Abstract In
More informationDetecting Behavior Propagation in BGP Trace Data Brian J. Premore Michael Liljenstam David Nicol
Detecting Behavior Propagation in BGP Trace Data Brian J. Premore Michael Liljenstam David Nicol Institute for Security Technology Studies, Dartmouth College 1 Motivation Is there a causal connection between
More informationThe Origin of BGP Duplicates
David Hauweele, Bruno Quoitin, Cristel Pelsser, Randy Bush To cite this version: David Hauweele, Bruno Quoitin, Cristel Pelsser, Randy Bush. The Origin of BGP Duplicates. CoRes 2016, May 2016, Bayonne,
More informationReducing BGP convergence time by fine tuning the MRAI timer on different topologies
Reducing BGP convergence time by fine tuning the MRAI timer on different topologies Pradeep, K and Alani, OYK Title Authors Type URL Published Date 2012 Reducing BGP convergence time by fine tuning the
More informationBGP Route Flap Damping Algorithms
BGP Route Flap Damping Algorithms Wei Shen and Ljiljana Trajković * Simon Fraser University Vancouver, British Columbia, Canada {wshen, ljilja}@cs.sfu.ca Keywords: Routing protocols, BGP, route flap damping,
More informationUSING SIMULATION TO UNDERSTAND DYNAMIC CONNECTIVITY AT THE CORE OF THE INTERNET
USING SIMULATION TO UNDERSTAND DYNAMIC CONNECTIVITY AT THE CORE OF THE INTERNET DAVID NICOL, BRIAN PREMORE MICHAEL LILJENSTAM ANDY OGIELSKI Dartmouth College Renesys Corporation Hanover, NH 03755 Hanover,
More informationFlooding Attacks by Exploiting Persistent Forwarding Loops
Flooding Attacks by Exploiting Persistent Forwarding Jianhong Xia, Lixin Gao, Teng Fei University of Massachusetts at Amherst {jxia, lgao, tfei}@ecs.umass.edu ABSTRACT In this paper, we present flooding
More informationDiscovering Interdomain Prefix Propagation using Active Probing
Discovering Interdomain Prefix Propagation using Active Probing lorenzo@ripe.net - colitti@dia.uniroma3.it ISMA 2006 WIT, San Diego, 10 May 2006 http://www.ripe.net 1 The problem ISMA 2006 WIT, San Diego,
More informationFinding a Needle in a Haystack: Pinpointing Significant BGP Routing Changes in an IP Network
Finding a Needle in a Haystack: Pinpointing Significant BGP Routing Changes in an IP Network Jian Wu (University of Michigan) Z. Morley Mao (University of Michigan) Jennifer Rexford (Princeton University)
More informationInferring the Origin of Routing Changes using Link Weights
Inferring the Origin of Routing Changes using Link Weights Mohit Lad, Ricardo Oliveira, Dan Massey, and Lixia Zhang Computer Science Department, University of California, Los Angeles CA 90095 Email: {mohit,rveloso,lixia}@cs.ucla.edu
More informationBGP Route Flap Damping Algorithms
BGP Route Flap Damping Algorithms Wei Steve Shen wshen@cs.sfu.ca Communication Networks Laboratory http://www.ensc.sfu.ca/cnl Simon Fraser University Roadmap Introduction to Route Flap Damping (RFD) ns-2
More informationImproving Packet Delivery Performance of BGP During Large-Scale Failures
Improving Packet Delivery Performance of BGP During Large-Scale Failures Amit Sahoo Dept. of Computer Science Univ. of Califonia, Davis Davis, CA 95616 Email:asahoo@ucdavis.edu Krishna Kant Intel Corporation
More informationSpeculative Route Invalidation to Improve BGP Convergence Delay under Large-Scale Failures
Speculative Route Invalidation to Improve BGP Convergence Delay under Large-Scale Failures Amit Sahoo Dept. of Computer Science Univ. of Califonia, Davis Davis, CA 95616 Email:asahoo@ucdavis.edu Krishna
More informationBGP Route Flap Damping Algorithms
BGP Route Flap Damping Algorithms Steve Shen and Ljiljana Trajković {wshen, ljilja}@cs.sfu.ca Communication Networks Laboratory http://www.ensc.sfu.ca/cnl Simon Fraser University Roadmap Introduction to
More informationQuantifying Path Exploration in the Internet
IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 17, NO. 2, APRIL 2009 445 Quantifying Path Exploration in the Internet Ricardo Oliveira, Member, IEEE, Beichuan Zhang, Dan Pei, and Lixia Zhang Abstract Previous
More informationBGP Path Exploration Damping (PED)
BGP Path Exploration Damping (PED) Mattia Rossi mrossi@swin.edu.au Centre for Advanced Internet Architectures (CAIA) Swinburne University of Technology Outline Introduction Motivation Path Exploration
More informationBGP Routing Stability of Popular Destinations
BGP Routing Stability of Popular Destinations Jennifer Rexford, Jia Wang, Zhen Xiao, and Yin Zhang AT&T Labs Research; Florham Park, NJ Abstract The Border Gateway Protocol (BGP) plays a crucial role in
More informationMulticast Routing Instabilities
Multicast Routing Instabilities Native multicast is critical for scaling the delivery of high-bandwidth data, such as audio and video, to multiple receivers. Until recently, however, multicast routing
More informationCollection and Characterization of BCNET BGP Traffic
Collection and Characterization of BCNET BGP Traffic Sukhchandan Lally lally@sfu.ca Communication Networks Laboratory http://www.ensc.sfu.ca/research/cnl School of Engineering Science Simon Fraser University
More informationDestination Reachability and BGP Convergence Time. Beichuan Zhang (UCLA) Dan Massey (Colorado State) Lixia Zhang (UCLA)
Destination Reachability and BGP Convergence Time Beichuan Zhang (UCLA) Dan Massey (Colorado State) Lixia Zhang (UCLA) Packet Delivery and Routing Dynamics The primary goal of routing is to deliver packets.
More informationK A I S T Department of Computer Science
Forensic Analysis of Automonous System Reachability DK Lee, Sue Moon, Taesang Choi, and Taesoo Jeong KAIST and ETRI CS/TR-26-257 May 26 K A I S T Department of Computer Science Forensic Analysis of Automonous
More informationQuantifying Path Exploration in the Internet
Quantifying Path Exploration in the Internet Ricardo Oliveira Beichuan Zhang Dan Pei Lixia Zhang {rveloso,lixia}@cs.ucla.edu bzhang@cs.arizona.edu peidan@research.att.com University of California, Los
More informationTaming BGP. An incremental approach to improving the dynamic properties of BGP. Geoff Huston. CAIA Seminar 18 August
Taming BGP An incremental approach to improving the dynamic properties of BGP Geoff Huston BGP is The inter-domain routing protocol for the Internet An instance of a Distance Vector Protocol with explicit
More informationCharacterization of BGP Recovery Time under Large-Scale Failures
Characterization of BGP Recovery Time under Large-Scale Failures Amit Sahoo Dept. of Computer Science Univ. of Califonia, Davis Davis, CA 95616 Email:asahoo@ucdavis.edu Krishna Kant Intel Corporation Hillsboro,OR
More informationA Visual Exploration Process for the Analysis of Internet Routing Data
A Visual Exploration Process for the Analysis of Internet Routing Data Soon Tee Teoh Kwan-Liu Ma S. Felix Wu Department of Computer Science University of California, Davis Abstract The Internet pervades
More informationTowards a Logic for Wide-Area Internet Routing
Towards a Logic for Wide-Area Internet Routing Nick Feamster and Hari Balakrishnan M.I.T. Computer Science and Artificial Intelligence Laboratory {feamster,hari}@lcs.mit.edu ; #, $. ', - -, * - ' * 4 *
More informationOn Reverse Engineering the Management Actions from Observed BGP Data
On Reverse Engineering the Management Actions from Observed BGP Data Shih Ming Tseng, S. Felix Wu University of California, Davis Email: {smtseng,sfwu}@ucdavis.edu Xiaoliang Zhao Verizon Communications
More informationA Measurement Study on the Impact of Routing Events on End-to-End Internet Path Performance
A Measurement Study on the Impact of Routing Events on End-to-End Internet Path Performance Feng Wang University of Mass., Amherst fewang@ecs.umass.edu Zhuoqing Morley Mao University of Michigan zmao@eecs.umich.edu
More informationMeasuring BGP. Geoff Huston. CAIA SEMINAR 31 May
Measuring BGP Geoff Huston BGP is An instance of the Bellman-Ford Distance Vector family of routing protocols And a relatively vanilla one at that The routing protocol used to support inter-domain routing
More informationIntroduction to IP Routing. Geoff Huston
Introduction to IP Routing Geoff Huston Routing How do packets get from A to B in the Internet? A Internet B Connectionless Forwarding Each router (switch) makes a LOCAL decision to forward the packet
More informationGlobal Routing Instabilities
Global Routing Instabilities during Code Red 2 and Nimda Worm Propagation Jim Cowie and Andy Ogielski Renesys Corporation 17 December 2001 www..com Abstract We review our recent discovery of the unexpected
More informationSimulating Internet Scale Topologies with Metarouting
Computer Science Technical Report Simulating Internet Scale Topologies with Metarouting Steve DiBenedetto, Andrew Stone, Michelle Strout, Dan Massey Department of Computer Science Colorado State University
More informationSoundness of Chilean Internet Routes
Soundness of Chilean Internet Routes Pablo Sepúlveda, Victor Ramiro, Tomás Barros NIC Chile Research Labs, Universidad de Chile Miraflores 222, Piso 14, 832-0198, Santiago, Chile. {psepulv,vramiro,tbarros}@niclabs.cl
More informationEULER Project Path-Vector Routing Stability Analysis
EULER Project Path-Vector Routing Stability Analysis Florin Coras, Albert Lopez, Albert Cabellos UPC Dimitri Papadimitriou Alcatel-Lucent Introduction BGP Inter-domain routing protocol used in the Internet
More informationThe Impact of Router Outages on the AS-Level Internet
The Impact of Router Outages on the AS-Level Internet Matthew Luckie* - University of Waikato Robert Beverly - Naval Postgraduate School *work started while at CAIDA, UC San Diego SIGCOMM 2017, August
More informationProtecting DNS from Routing Attacks -Two Alternative Anycast Implementations
Protecting DNS from Routing Attacks -Two Alternative Anycast Implementations Boran Qian StudentID 317715 Abstract The Domain Names System (DNS) is an important role of internet infrastructure and supporting
More informationClassification of BGP Anomalies Using Decision Trees and Fuzzy Rough Sets
Classification of BGP Anomalies Using Decision Trees and Fuzzy Rough Sets Yan Li, Hong-Jie Xing, Qiang Hua, Xi-Zhao Wang, Prerna Batta, Soroush Haeri, and Ljiljana Trajković Hebei University, Baoding,
More informationMeasuring BGP Pass-Through Times
Measuring BGP Pass-Through Times Anja Feldmann 1, Hongwei Kong 2, Olaf Maennel 1, and Alexander Tudor 3 1 Technische Universität München, Germany {anja,olafm}@net.in.tum.de 2 Agilent Labs, Beijing, China
More informationComprehensive Solution for Anomaly-free BGP
Comprehensive Solution for Anomaly-free BGP Ravi Musunuri, Jorge A. Cobb Department of Computer Science, The University of Texas at Dallas, Richardson, TX-7083-0688 musunuri, cobb @utdallas.edu Abstract.
More informationFeedback Based Routing
Feedback Based Routing Dapeng Zhu dapengz@cs.stanford.edu Mark Gritter mgritter@cs.stanford.edu Department of Computer Science Stanford University Stanford, CA 94305 David R. Cheriton cheriton@cs.stanford.edu
More informationTHE INTERNET S inter-domain routing protocol, the
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 28, NO. 8, OCTOBER 2010 1271 A Technique for Reducing BGP Update Announcements through Path Exploration Damping Geoff Huston, Mattia Rossi, and Grenville
More informationThe Internet now-a-days has become indispensable to each and everyone. It is vulnerable to
CHAPTER 5 NETWORK AVAILABILITY IN INTERDOMAIN ROUTING The Internet now-a-days has become indispensable to each and everyone. It is vulnerable to node failures, link failures, and fluctuations due to many
More informationTowards Localizing Root Causes of BGP Dynamics
Towards Localizing Root Causes of BGP Dynamics Matthew Caesar, Lakshminarayanan Subramanian, Randy H. Katz {mccaesar,lakme,randy}@cs.berkeley.edu Abstract Today, we lack a clear understanding of the dynamics
More informationRouting, Routing Algorithms & Protocols
Routing, Routing Algorithms & Protocols Computer Networks Lecture 6 http://goo.gl/pze5o8 Circuit-Switched and Packet-Switched WANs 2 Circuit-Switched Networks Older (evolved from telephone networks), a
More informationPreventing the unnecessary propagation of BGP withdraws
Preventing the unnecessary propagation of BGP withdraws V. Van den Schrieck, P. François, C. Pelsser, O.Bonaventure http://inl.info.ucl.ac.be Networking 2009, May 13th Agenda Introduction Presentation
More informationQuantifying the BGP routes diversity inside a tier-1 network
Quantifying the BGP routes diversity inside a tier-1 network Steve Uhlig, Sébastien Tandel Department of Computing Science and Engineering Université catholique de Louvain, Louvain-la-neuve, B-1348, Belgium
More informationForensic Analysis of Autonomous System Reachability
Forensic Analysis of Autonomous System Reachability DK Lee, Sue Moon Division of Computer Science, KAIST {dklee, sbmoon}@an.kaist.ac.kr Taesang Choi, Taesoo Jeong ETRI {choits, tsjeong}@etri.re.kr ABSTRACT
More informationBGP Anomaly Detection. Bahaa Al-Musawi PhD candidate Supervisors: Dr. Philip Branch and Prof. Grenville Armitage.
BGP Anomaly Detection Bahaa Al-Musawi PhD candidate Supervisors: Dr. Philip Branch and Prof. Grenville Armitage balmusawi@swin.edu.au Centre for Advanced Internet Architectures (CAIA) Swinburne University
More informationGLOBAL INTERNET ROUTING FORENSICS Validation of BGP Paths using ICMP Traceback
Chapter 14 GLOBAL INTERNET ROUTING FORENSICS Validation of BGP Paths using ICMP Traceback Eunjong Kim, Dan Massey and Indrajit Ray Abstract The Border Gateway Protocol (BGP), the Internet's global routing
More informationImproving BGP Convergence Through Consistency Assertions
Improving BGP Convergence Through Consistency Assertions Dan Pei, Xiaoliang Zhao, Lan Wang, Dan Massey, Allison Mankin, S. Felix Wu, Lixia Zhang 1 Abstract This paper presents a new mechanism for improving
More informationCS4450. Computer Networks: Architecture and Protocols. Lecture 15 BGP. Spring 2018 Rachit Agarwal
CS4450 Computer Networks: Architecture and Protocols Lecture 15 BGP Spring 2018 Rachit Agarwal Autonomous System (AS) or Domain Region of a network under a single administrative entity Border Routers Interior
More informationInterdomain routing CSCI 466: Networks Keith Vertanen Fall 2011
Interdomain routing CSCI 466: Networks Keith Vertanen Fall 2011 Overview Business relationships between ASes Interdomain routing using BGP Advertisements Routing policy Integration with intradomain routing
More informationEvaluation of Prefix Hijacking Impact Based on Hinge-Transmit Property of BGP Routing System
Evaluation of Prefix Hijacking Impact Based on Hinge-Transmit Property of BGP Routing System Evaluation of Prefix Hijacking Impact Based on Hinge-Transmit Property of BGP Routing System School of Computer,
More informationEffect of Minimal Route Advertisement Interval Timers on Border Gateway Protocol Convergence
Effect of Minimal Route Advertisement Interval Timers on Border Gateway Protocol Convergence Rajvir Gill rajvirg@sfu.ca Communication Networks Laboratory http://www.ensc.sfu.ca/research/cnl School of Engineering
More informationKey words: TCP/IP, IGP, OSPF Routing protocols, MRC, MRC System.
www.ijecs.in International Journal Of Engineering And Computer Science ISSN:2319-7242 Volume2 Issue 8 August, 2013 Page No. 2628-2634 Multiple Routing Configurations For Fast Ip Network Recovery Sk.Meeravali
More informationTop-Down Network Design
Top-Down Network Design Chapter Seven Selecting Switching and Routing Protocols Original slides by Cisco Press & Priscilla Oppenheimer Selection Criteria for Switching and Routing Protocols Network traffic
More informationInternet Routing Dynamics
Internet Routing Dynamics CS589 Lecture 2 Z. Morley Mao Jan 11, 2004 Z. Morley Mao, Winter 2005, CS589 1 Two types of Internet Routing Protocols Internet consists of roughly 19,000 Autonomous Systems What
More informationBGP Route Propagation between Neighboring Domains
BGP Route Propagation between Neighboring Domains Renata Teixeira 1, Steve Uhlig 2, and Christophe Diot 3 1 Univ. Pierre et Marie Curie, LIP6-CNRS, renata.teixeira@lip6.fr 2 Delft University of Technology
More informationA Longitudinal Study of BGP MOAS Prefixes
A Longitudinal Study of BGP MOAS Prefixes Quentin Jacquemart Eurecom Sophia Antipolis Guillaume Urvoy-Keller Univ. Nice Sophia Antipolis, CNRS, I3S, UMR 7271, 06900 Sophia Antipolis Ernst Biersack Eurecom
More informationOn characterizing BGP routing table growth
University of Massachusetts Amherst From the SelectedWorks of Lixin Gao 00 On characterizing BGP routing table growth T Bu LX Gao D Towsley Available at: https://works.bepress.com/lixin_gao/66/ On Characterizing
More informationPerformance Evaluation of Border Gateway Protocol with Route Flap Damping and Routing Policies
Performance Evaluation of Border Gateway Protocol with Route Flap Damping and Routing Policies Ravinder Paul rpa28@sfu.ca Communication Networks Laboratory http://www.sfu.ca/~ljilja/cnl School of Engineering
More informationQuantifying Path Exploration in the Internet
Quantifying Path Exploration in the Internet Ricardo Oliveira rveloso@cs.ucla.edu Beichuan Zhang bzhang@cs.arizona.edu Rafit Izhak-Ratzin rafiti@cs.ucla.edu Lixia Zhang lixia@cs.ucla.edu Dan Pei peidan@research.att.com
More informationTHE Internet is an interconnection of separately administered
SPRINT ATL RESEARCH REPORT RR3-ATL-51677 - MAY 23 1 A Study of the Impact of BGP Dynamics on Intra-Domain Sharad Agarwal Chen-Nee Chuah Supratik Bhattacharyya Christophe Diot CS Division, ECE Department,
More informationUnderstanding Transit-Edge Routing Separation: Analysis and Characterization
Understanding Transit-Edge Routing Separation: Analysis and Characterization Kunpeng Liu a, Bijan Jabbari a, Stefano Secci b a Communications and Networking Lab., George Mason University, VA. b LIP6, Pierre
More informationInternet Routing : Fundamentals of Computer Networks Bill Nace
Internet Routing 14-740: Fundamentals of Computer Networks Bill Nace Material from Computer Networking: A Top Down Approach, 6 th edition. J.F. Kurose and K.W. Ross Looking Ahead Lab #2 just due Quiz #2
More informationIntroduction to BGP ISP/IXP Workshops
Introduction to BGP ISP/IXP Workshops 1 Border Gateway Protocol Routing Protocol used to exchange routing information between networks exterior gateway protocol RFC1771 work in progress to update draft-ietf-idr-bgp4-18.txt
More informationA Way to Implement BGP with Geographic Information
International Journal of Electronics Engineering, 2 (2), 2010, pp. 349 353 A Way to Implement BGP with Geographic Information Neha Mangla 1 and R.K. Khola 2 1 Department of Computer Science, Suresh Gyanvihar
More informationActive BGP Probing. Lorenzo Colitti. Roma Tre University RIPE NCC
Active BGP Probing Lorenzo Colitti Roma Tre University RIPE NCC Lorenzo Colitti. RIPE 50, 5 May 2005. colitti@dia.uniroma3.it lorenzo@ripe.net 1 Agenda Our techniques Primitives Applications Results Operational
More informationInter-Domain Routing: BGP
Inter-Domain Routing: BGP Richard T. B. Ma School of Computing National University of Singapore CS 3103: Compute Networks and Protocols Inter-Domain Routing Internet is a network of networks Hierarchy
More informationDetection of Invalid Routing Announcement in the Internet Λ
Detection of Invalid Routing Announcement in the Internet Λ Xiaoliang Zhao, Dan Pei, Lan Wang, Dan Massey, Allison Mankin, S. Felix Wu,Lixia Zhang y Abstract Network measurement has shown that a specific
More informationSecuring BGP Networks using Consistent Check Algorithm
Securing BGP Networks using Consistent Check Algorithm C. K. Man, K.Y. Wong, and K. H. Yeung Abstract The Border Gateway Protocol (BGP) is the critical routing protocol in the Internet infrastructure.
More informationUCLA UCLA Previously Published Works
UCLA UCLA Previously Published Works Title IPv address allocation and the BGP routing table evolution Permalink https://escholarship.org/uc/item/8b6dw Journal Computer Communication Review, () ISSN 6-8
More informationImplementation of BGP in a Network Simulator
Implementation of BGP in a Network Simulator Tony Dongliang Feng Rob Ballantyne Ljiljana Trajković Communication Networks Laboratory http://www.ensc.sfu.ca/cnl Simon Fraser University Road map Introduction
More informationGuidelines for Interdomain Traffic Engineering
Guidelines for Interdomain Traffic Engineering Nick Feamster Jay Borkenhagen Jennifer Rexford Laboratory for Computer Science AT&T IP Services Internet and Networking Systems Massachusetts Institute of
More informationBGP Routing and BGP Policy. BGP Routing. Agenda. BGP Routing Information Base. L47 - BGP Routing. L47 - BGP Routing
BGP Routing and BGP Policy BGP Routing The BGP Routing Principles and Route Decisions based on AS-Path in a simple topology of AS s routing policy is reduced to a minimal function demonstrated in example
More informationLongitudinal Study of BGP Monitor Session Failures
Longitudinal Study of BGP Monitor Session Failures ABSTRACT Pei-chun Cheng Department of Computer Science University of California, Los Angeles pccheng@cs.ucla.edu Beichuan Zhang Department of Computer
More informationBGP Security via Enhancements of Existing Practices
IEEE International Conference on Communications 2009 1 BGP Security via Enhancements of Existing Practices Xiaoliang Zhao, David T. Kao * Abstract Border Gateway Protocol (BGP) is the de-facto inter-domain
More informationBGP Convergence Delay under Large-Scale Failures: Characterization and Solutions
1 BGP Convergence Delay under Large-Scale Failures: Characterization and Solutions Amit Sahoo, Krishna Kant, Member, IEEE, and Prasant Mohapatra, Senior Member, IEEE Abstract Border gateway protocol (BGP)
More informationSaaS Providers. ThousandEyes for. Summary
USE CASE ThousandEyes for SaaS Providers Summary With Software-as-a-Service (SaaS) applications rapidly replacing onpremise solutions, the onus of ensuring a great user experience for these applications
More informationOn Detection of Anomalous Routing Dynamics in BGP
On Detection of Anomalous Routing Dynamics in BGP Ke Zhang1, Amy Yen 1, Xiaoliang Zhao3, Dan Massey 3, S. Felix Wu1, and Lixia Zhang 4 1 University of California, Davis, CA, U.S.A {kezhang,ahyen,sfwu}@ucdavis.edu
More informationCS 557 BGP Convergence
CS 557 BGP Convergence Improved BGP Convergence via Ghost Flushing Bremler-Barr, fek, Schwarz, 2003 BGP-RCN: Improving Convergence Through Root Cause Notification Pei, zuma, Massey, Zhang, 2005 Spring
More informationRouting the Internet in Geoff Huston APNIC March 2007
Routing the Internet in 2006 Geoff Huston APNIC March 2007 IPv4 in 2006 Total Advertised BGP Prefixes BGP Table Growth 210000 200000 190000 BGP FIB Entries 180000 2006 170000 2005 160000 150000 Jan Feb
More informationCyclops: The AS-level Connectivity Observatory
Cyclops: The AS-level Connectivity Observatory Ying-Ju Chi, Ricardo Oliveira, and Lixia Zhang University of California, Los Angeles {yjchi,rveloso,lixia}@cs.ucla.edu ABSTRACT In this paper we present Cyclops,
More informationEarly Detection of BGP Instabilities Resulting from Internet Worm Attacks
Early Detection of BGP Instabilities Resulting from Internet Worm Attacks S. Deshpande ECSE Department, Rensselaer Polytechnic Institute Troy, NY 28 M. Thottan Center for Networking Research, Bell aboratories
More informationLOUP: The Principles and Practice of Intra-Domain Route Dissemination. Nikola Gvozdiev, Brad Karp, Mark Handley
LOUP: The Principles and Practice of Intra-Domain Route Dissemination Nikola Gvozdiev, Brad Karp, Mark Handley The Rising Tide of Reachability Expectations Internet users expect any-to-any reachability:
More informationThe Impact of BGP Dynamics on Intra-Domain Traffic
The Impact of BGP Dynamics on Intra-Domain Traffic Sharad Agarwal Chen-Nee Chuah Supratik Bhattacharyya Christophe Diot CS Division, ECE Department, Sprint ATL, Intel Research, University of California,
More informationbgpand - Architecting a modular BGP4 Attack & Anomalies Detection Platform
bgpand - Architecting a modular BGP4 Attack & Anomalies Detection Platform Mayank Bhatnagar TechMahindra Limited, SDF B-1, NSEZ, Noida-201305, India E-mail : mayank.bhatnagar2@techmahindra.com Abstract
More informationOn the Characteristics of BGP Routes
On the Characteristics of BGP Routes Julien Clément and Kwan-Wu Chin School of Electrical, Computer and Telecommunications Engineering University of Wollongong Northfields Ave, NSW, Australia 2522 {jyc157,
More informationAPT Incremental Deployment
APT Incremental Deployment Dan Jen, Michael Meisel, Daniel Massey, Lan Wang, Beichuan Zhang, Lixia Zhang http://www.cs.ucla.edu/~meisel/draft-apt-incremental-00.txt 1 Why This Talk Incrememtal deployability
More informationPHAS: A Prefix Hijack Alert System
PHAS: A Prefix Hijack Alert System Mohit Lad mohit@cs.ucla.edu Yiguo Wu yiguowu@cs.ucla.edu Dan Massey massey@cs.colostate.edu Beichuan Zhang bzhang@cs.arizona.edu Dan Pei peidan@research.att.com Lixia
More informationUnderstanding Transit-Edge Routing Separation: Analysis and Characterization
Understanding Transit-Edge Routing Separation: Analysis and Characterization Kunpeng Liu a, Bijan Jabbari a, Stefano Secci b a Communications and Networking Lab., George Mason University, VA. b LIP6, Pierre
More information